1908] SHULLMENDELIAN INHERITANCE 107 



delian, but still leave a doubt on account of the obvious derangement 

 produced by ill-treatment. 



In the summer of 1907 the seeds were sown at the proper time 

 (April 30) and the Mendelian character of the branching habit was 

 fully established by the following facts: (a) Reciprocal crosses 

 (06356 and 06357), fully controlled, between two strongly branched 

 specimens belonging to the second of the two DR families mentioned 

 in 1906 (namely, no. 05150) gave 65 fully branched individuals in 

 the one family, 13 in the other family, and no unbranched specimens 

 in either, thus showing that at least one of the chosen parents was an 

 extracted dominant (DD). The number of branches in both these 

 reciprocal families ranged from 10 to 25. (b) Properly controlled 

 pollinations w re made between two specimens of the last family (05145) 

 mentioned for 1906, supposedly a recessive family. The pistil- 

 parent had two strong branches, one of which showed an abnormal 

 bifission. The pollen-parent had one strong lateral branch w r hich ex- 

 ceeded the main stem. The cross was made between two heads borne 

 by lateral branches. Of the 22 offspring produced (06353), 19 were 

 wholly unbranched, two had one small branch, each near the base of 

 the stem, and one had two small branches similarly located. This 

 demonstrated that both parents were pure recessives and showed that 

 the branches produced the preceding year had no hereditary signifi- 

 cance, (c) Another sowing (06352) was made of unguarded seed from 

 the same pistil-parent as the last. The 91 offspring consisted of 67 

 individuals having less than 7 branches each and 24 having more than 7 

 branches each, the latter group having an average of about 16 branches 

 and the former group an average of less than two branches (fig. j). 

 As the branched type does not usually have more than 4 or 5 vacant 

 axils above the highest branch, one individual having 8 branches 

 and 10 vacant axils probably belongs with the unbranched class. 

 The result in this family shows that about one -fourth of the pollen 

 received by the unguarded mother came from branched individuals, 

 and as this fact was made obvious in the first generation it gives further 

 proof of the dominance of the branching habit over the unbranched. 



Furthermore, it illustrates well how impossible it would be to 

 discover Mendelian ratios in cross-fertilizing species or varieties with- 

 out careful control of fertilization. This requirement accounts for 



